Question

1.

The followings are the performance measures generated from multiple runs of the k-Means model on the same dataset with different configurations. Which one presents the best k-Means model.

Select one:

a. SSE: 303; Davies-Bouldin index: 0.29

b. SSE: 706: Davies-Bouldin index: 0.85

c. SSE: 866; Davies-Bouldin index: 0.87

2.

You have to specify which of the following parameters before the run of the DBSCAN algorithm.

Select one:

a. epsilon

b. epsilon and MinPoints

c. Number of clusters (k)

d. MinPoints

Answer 1

1)​​​​​​ Option is (a)

a. SSE: 303; Davies-Buldin index: 0.29

SSE is defined as the Squared sum of the distance between the centroid and all of the members in the cluster. For a good cluster, Intra-class similarity should be high, which means as low SSE as possible.

Davies-Bouldin Index: Lower Davies-Bouldin Index clusters means that the algorithm is having a better quality.

Based on both of these option A has lowest of both out of three. Hence, "A" is answer.

Reasons for incorrect answer

• Option b and c is having higher index than the option a.

2)

option b is correct.

epsilon and minpoints are to be specified before the run of DBSCAN algorithm. Two focuses are been viewed as neighbours when the seperation between two of the focuses are underneath the limit epsilon and also the minimum number of neighbours of the given point.This algorithm works with these two parameters. The epsilon specifies how close the points must be for each other for the part of a cluster. The minPoints is the minimum number of the points for a dense region.

Reason for incorrect options-

• Epsilon is required but only epsilon is not enough.
• The number of clusters is not required to be mentioned because here in the algorithm as the modell may be trained on unlabelled data points. The k will be discovered by the algorithm.
• Only minpoint is not enough to run the algorithm.

Please give an upvote if you are satisfied. Else leave a comment. We love helping you